Apple M5 Chip's Dual-Use Design Will Power Future Macs and AI Servers

This sounds like the theorized "Double Ultra" chips that would have been two M1 Ultra Chipsets stacked on top of each other, connected at the beltline interface between two Mx Maxs.

It's a bit more sophisticated than that. Right now a SoC has all of its components on a monolithic die - the CPU and GPU cores, the NPU, memory controllers, built in USB/Thunderbolt controllers, Secure Enclave, etc. What SoIC allows for is all of those SoC components to be split out and fabricated separately but then integrated back with each other with resulting performance that's actually better than a monolithic SoC die.

So instead of an Ultra being two Max chips being stitched together it could be a large CPU tile joined to a large GPU tile with smaller controller component tiles added on. All of the M series chips would be built this way, not just the Ultra. This would have the side effect of increased yields by making smaller and less complex chiplet dies, and allow Apple to focus limited bleeding edge process node capacity on important sections of the tiles like the CPU and GPU and have controller tiles manufactured on older but still efficient process nodes.

Is Apple getting back into the server market?

They never left. The M1 Mac mini for $275 lightly used runs circles around the Xserves ?

Is Apple getting back into the server market?

"Server" covers a lot of different uses. I'm sure the focus will be on systems designed to drive AI and high-performance computing services - a booming requirement - rather than file sharing/email/web-hosting. ...but there is already stiff competition there, including power efficient ARM based systems. The competition will include AWS, who have their own ARM-based processors, and Nvidia's Grace Hopper ('https://www.nvidia.com/en-gb/data-center/grace-hopper-superchip/') chip (which is a good illustration of where multi-chip technology might come in).

They never left. The M1 Mac mini for $275 lightly used runs circles around the Xserves

There's a difference between hardware that can be used as a server - the Mac Mini - and a system designed specifically for datacentre applications - like the XServe. Datacentre systems have features nothing to do with the CPU, like

* being designed from the ground up for high density rack mounting (including layout and cooling)
* dual redundant power supplies (PSU failure is a onec-in-a-blue-moon a single personal computer, in a datacentre with hundreds of the things, it must be Thursday!)
* hot-swap/RAID disc drives (same logic as above)
* lights-out power management (for those times when you need to push the power button - again, same logic)

When the PPC XServe was released it had several Unique Selling Points: the PPC still had a claim to be better than Intel, Linux hadn't gained widespread acceptance - wheras Mac OS was certified as Unix which gave it cred (and let it run a lot of standard software) and the competition was proprietary stuff like Novel Netware and commercial Unix systems which (a) often offered shoddy (& expensive) support for file sharing and mail on Mac networks and (b) had expensive per-seat licensing plans.

When the Intel XServe was discontinued, it really had little left to distinguish it from cheap & cheerful Intel servers using the same processors. Customers that still used Windows Server or commercial Unix did so because they weren't inclined to change - and personal Macs had adapted to work in such environments (e.g. shifting from AppleTalk and Apple file sharing protocol to TCP/IP and SMB) and the industry was moving towards Linux and open web/internet protocols and open source server software anyway. An XServe could do the job perfectly well - but so could a generic Intel or AMD server running Linux, for less money. The Mac's primary advantage was its GUI and cool industrial design - which doesn't count for much in a server room running software that techies can configure using text files. Take the GUI and user-centric Apps away and it was an x86 system running BSD - Open Group certification still counts for something, but Linux is now the de-facto standard setter. Even Apple failed to eat their own dog food and filled their data centres with x86 black boxes.

What's changed now is that Apple are in the CPU and GPU business like never before so they have the opportunity to produce something distinctive for AI and HPC services. The M1/2/3 series don't really cut the mustard, but Apple have the building blocks to make something special - if there is a big enough market to justify the development, which there probably isn't for Mac Pros.

However, the main market for future Apple server kit could be Apple themselves - to run services that not only make money but sell iPhones and Macs - if they don't want to keep contracting their AI out to third parties.

M5 Mulitronic - what could go wrong.?

Is Apple getting back into the server market?

Here's hoping the M5 will be based on the 2nm architecture with TSMC Gate All Around structure.

Also hoping by this time 16GB intergrated RAM will be the standard RAM size for the Mac.